A laser ultrasonic intelligent inspection method for metal surface defects based on digital twin model

To enhance the efficiency and intelligence of laser ultrasonic detection of metal surface defects, a detection method based on digital twins is proposed. Given the scarcity of defect samples and the difficulty of obtaining labeled data in practical applications, the finite element method is used to construct a laser ultrasonic metal defect detection model, effectively generating twin data for defect detection. With the help of generative adversarial networks, the distribution difference between twin data and real data is minimized. Additionally, by using continuous wavelet transform, one-dimensional ultrasonic signals are converted into two-dimensional time-frequency images for detection input. In terms of detection algorithm development, a residual network with an optimized global attention mechanism is introduced to ensure real-time synchronization and updates between the digital twin model and the actual laser ultrasonic detection equipment. Finally, experimental validation of the proposed method shows that it effectively addresses the issue of small sample sizes in defect detection and achieves high detection accuracy. Moreover, comparative experiments on crack defect migration tasks under different settings demonstrate the method's good generalization ability.